Transparent object surface defect detection method based on transmission structured light

Abstract

The invention discloses a transparent object surface defect detection method based on transmission structured light. The method sequentially comprises the following steps: establishing a measurement system; generating a stripe image; a vertical deformation fringe pattern and a horizontal deformation fringe pattern are collected and demodulated, and a vertical unwrapping phase and a horizontal unwrapping phase of the transparent glass plate to be detected are obtained respectively; according to the vertical unwrapping phase and the horizontal unwrapping phase, the surface scratch area and the flatness error of the transparent glass plate to be detected are obtained; judging whether the transparent glass plate to be detected meets the requirements or not according to the two indexes of the surface scratch area and the flatness error: when the two indexes meet the respective requirements, judging that the transparent glass plate to be detected is qualified and meets the requirements. Under the condition that calibration is not needed, the surface shape error is detected through the transmission-type phase deflection measurement system, the detection rate is increased, the system is simplified, and comprehensive detection of the surface shape error and the surface defect is achieved.

Description

Technical Field

[0001] The present invention relates to the field of transparent object defect detection, and specifically to a method for detecting surface defects of transparent objects based on transmissive structured light. Background Art

[0002] The quality indicators of precision optical components mainly include optical surface form error and surface defects. The surface form error refers to the shape deviation of the optical device surface, including the degree to which the geometric forms of the surfaces of optical components such as lenses and mirrors do not conform to the ideal form. The surface form error will have an important impact on the imaging quality of the optical system, and may cause problems such as aberration, defocus, and blurring, reducing the resolution and performance of the system. Therefore, in the design and manufacturing process of optical devices, it is necessary to strictly control and compensate for the surface form error to ensure that the optical system has the required accuracy and performance. Surface defects refer to fine features such as pits, scratches, and ripples on the mirror surface. These defects will not only cause the transmitted light beam to scatter, result in energy loss, harmful glare, and diffraction fringes, but also cause film layer damage and laser damage to the optical component.

[0003] The existing methods for detecting surface defects of transparent objects can be classified into two categories according to the optical characteristics they rely on: one is the detection method guided by the reflection optical path, and the other is the detection method guided by the transmission optical path.

[0004] The detection method guided by the reflection optical path utilizes the reflection of the incident light on the front surface to be measured without considering its propagation inside the object to be measured. This type of method avoids the complex optical path analysis and design process. The existing reflection optical path measurement techniques for the traditional detection method guided by the reflection optical path are as follows:

[0005] In the literature "Zhang H Y, Wang Z H, Fu H Y. Automatic scratch detector for optical surface[J]. Optics Express, 2019, 27(15): 20910 - 20927.", the dark field imaging method is adopted, and the difference in the directionality of the outgoing light reflected by the defect area and the non-damaged area is used to identify the defect. This method relies on the optical path deviation caused by surface defects, but the deviation generated by tiny defects is too small and is easily submerged by background noise. Therefore, dark field imaging is difficult to be applicable to the detection of tiny defects.

[0006] In the literature "Schneider V M, Mlejnek M, Gahagan K T. Fast detection of single-sided diffracted defects in display glass[J]. Measurement, 2009, 42(4): 638-644.", the single-sided diffraction strategy method was adopted to complete the detection of surface defects. However, these methods ignore the parasitic reflections brought by the outgoing light reflected by the back surface of the object to be measured, thus introducing additional detection errors.

[0007] The detection method guided by the transmission optical path uses the transmission light field formed by the incident light passing through the transparent object to achieve surface defect detection. Since the incident light needs to pass through the transparent object through two or more refractions on the front and back surfaces of the transparent object, the measured transparent medium has a relatively low reflectivity due to its physical property of high transmittance. When the light undergoes a double reflection process (that is, the light is first reflected on the back surface, then reflected again on the front surface, and finally transmitted out of the measured medium), its light intensity is greatly weakened, almost reaching a negligible level. The transmission structured light system precisely makes use of this physical property to effectively avoid the parasitic fringe problem commonly existing in the reflection optical system. The existing transmission optical path measurement techniques are specifically as follows:

[0008] In the literature "Method for detecting surface defects of transparent objects based on polarization transmission structured light[J]. Acta Optica Sinica, 2021, 41(18): 1812002", the polarization imaging method was adopted. A polarizer was added to the system, and by adjusting the angle of the light transmission axis of the polarizer, the distribution map of dust and the common distribution map of defects and dust were obtained. The defect distribution result was obtained through operations such as binarization, intersection taking, addition and subtraction.

[0009] It can be seen from the above methods that although the detection method guided by the transmission optical path can effectively avoid the parasitic fringe problem commonly existing in the reflection optical system, it does not take into account the surface shape error. Therefore, there is an urgent need for a comprehensive and effective method to detect the surface quality of optical elements. Summary of the Invention

[0010] Aiming at the deficiencies of the prior art, the technical problem to be solved by the present invention is to provide a method for detecting surface defects of transparent objects based on transmission structured light.

[0011] The technical solution of the present invention to solve the above technical problem is to provide a method for detecting surface defects of transparent objects based on transmission structured light, characterized in that the method includes the following steps:

[0012] Step 1: Build a measurement system;

[0013] The measurement system includes a display screen, a transparent glass plate to be measured, a camera, and a computer; the computer is communicatively connected to the display screen and the camera respectively; the display screen is used to display the fringe image generated by the computer; the camera is used to collect the deformed fringe pattern transmitted by the transparent glass plate to be measured;

[0014] The optical centers of the display screen, the camera, and the transparent glass plate to be measured are located on the same optical axis, and the transparent glass plate to be measured is located between the display screen and the camera; the deformed fringe pattern transmitted by the transparent glass plate to be measured can be clearly imaged within the measurement field of the camera;

[0015] Step 2: Generate a fringe image;

[0016] Step 3: Collect the vertical deformed fringe pattern and the horizontal deformed fringe pattern;

[0017] Step 4: Demodulate the vertical deformed fringe pattern and the horizontal deformed fringe pattern to obtain the vertical unfolded phase and the horizontal unfolded phase of the transparent glass plate to be measured respectively;

[0018] Step 5: Based on the vertical unfolded phase and the horizontal unfolded phase obtained in Step 4, obtain the surface scratch area and the flatness error of the transparent glass plate to be measured;

[0019] Step 6: Comprehensively consider the two indicators of the surface scratch area and the flatness error obtained in Step 5 to determine whether the transparent glass plate to be measured meets the requirements: when both indicators meet their respective requirements, it is determined that the transparent glass plate to be measured is qualified and meets the requirements.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0021] (1) The defect detection method of the present invention not only considers the defect of the scratch type, but also simultaneously considers the influence of the surface shape error on the surface quality.

[0022] (2) Without calibration, the present invention detects the surface shape error through a transmissive phase deflectometry system, improves the detection rate and simplifies the system. The deformed fringes passing through the transparent object are collected by an industrial camera, and the phase data is obtained by using the phase shift method and the three - fringe selection method to analyze the phase mutation for defect localization. Finally, by comparing the unfolded phase with the standard plane phase and calculating the root - mean - square error, it is determined whether the flatness meets the standard, realizing the comprehensive detection of the surface shape error and the surface defect.

[0023] (3) By counting the number of defective pixels extracted from the defect, the present invention can estimate the defect area, providing data support for the defect detection of the product.

[0024] (4) In an oblique-axis reflection optical system, after light is refracted by the front surface of a transparent medium, it is reflected again at the rear surface, thereby forming parasitic reflected light similar to the reflected light on the front surface. Defects on the front surface of this medium appear as double images in the modulation image, resulting in stripe superposition and confusion in the reflected optical system image. In the scope of structured light reflection technology, such a reflection phenomenon is specifically defined as parasitic stripes. Utilizing the physical property that the light intensity is greatly weakened during the process of light passing through a transmission system and undergoing two reflections, the problem of parasitic stripes commonly existing in reflection optical systems can be effectively avoided.

[0025] (5) The detection of surface defects of transparent objects based on phase deflectometry utilizes the light path refraction effect. When light passes through a transparent object, the defective surface will cause the light path to change. If the surface is smooth, the light will only undergo parallel displacement after refraction and still reach the bright area; if the surface has defects, the light will reach the dark area after refraction, forming bright spots. Defects are manifested as sudden changes in light intensity in the camera image. By analyzing the phase mutations in the unwrapped phase diagram, the defect positions can be determined. Brief Description of the Drawings

[0026] Figure 1 is a schematic diagram of the overall process of the present invention;

[0027] Figure 2 is a schematic diagram of the structure of the measurement system of the present invention;

[0028] Figure 3 are four green horizontal sinusoidal straight stripes with 64 fringes obtained in Embodiment 1 of the present invention;

[0029] Figure 4 are three groups of horizontal folded phase diagrams with different frequencies in Embodiment 1 of the present invention;

[0030] Figure 5 is the horizontal unwrapped phase diagram in Embodiment 1 of the present invention;

[0031] Figure 6 is the phase gradient amplitude diagram in Embodiment 1 of the present invention;

[0032] Figure 7 is the scratch area diagram in Embodiment 1 of the present invention.

[0033] In the figure, there are display screen 1, transparent glass flat plate to be measured 2, camera 3, and computer 4. Detailed Embodiment

[0034] The following gives specific embodiments of the present invention. The specific embodiments are only used to further illustrate the present invention in detail and do not limit the protection scope of the present invention.

[0035] The present invention provides a method for detecting surface defects of transparent objects based on transmissive structured light (hereinafter referred to as the method), which is characterized in that the method comprises the following steps:

[0036] Step 1: Set up a measurement system;

[0037] The measurement system includes a display screen 1, a transparent glass plate 2 to be measured, a camera 3, and a computer 4; the computer 4 is communicatively connected to the display screen 1 and the camera 3 respectively, controls the camera 3 and the display screen 1, and stores, displays, and processes the acquired images to obtain corresponding measurement results; the display screen 1 is used to display the fringe images generated by the computer 4; the camera 3 is used to collect the deformed fringe images transmitted by the transparent glass plate 2 to be measured.

[0038] Adjust the positional relationship among the display screen 1, the camera 3, and the transparent glass plate 2 to be measured so that the optical centers of the display screen 1, the camera 3, and the transparent glass plate 2 to be measured are on the same optical axis, and the transparent glass plate 2 to be measured is located between the display screen 1 and the camera 3; the deformed fringe images transmitted by the transparent glass plate 2 to be measured can be clearly imaged within the measurement field of the camera 3.

[0039] Preferably, in step 1, the computer 4 is communicatively connected to the display screen 1 through a VGA interface and to the camera 3 through a gigabit network interface.

[0040] Preferably, in step 1, the display screen 1 uses an LCD display screen; the camera 3 uses a CCD camera. Software for generating fringe images and image processing (MATLAB software in this embodiment), as well as software for controlling the camera 3 to collect and store images, are installed in the computer 4.

[0041] Step 2: Generate fringe images;

[0042] Preferably, in step 2, the computer 4 is used to generate three groups of vertical sinusoidal straight fringes and three groups of horizontal sinusoidal straight fringes that meet the requirements of four-step phase shifting and the selection of the best three fringes respectively. The number of fringes in the three groups of vertical sinusoidal straight fringes and the three groups of horizontal sinusoidal straight fringes both meet the selection of the best number of three fringes, and each group of vertical sinusoidal straight fringes and each group of horizontal sinusoidal straight fringes contain four fringe images with a 90° phase shift from each other.

[0043] Preferably, in step 2, the fringe images are generated by the software for generating fringe images and image processing installed in the computer 4.

[0044] Preferably, in step 2, the method of generating fringes by computer 4 is a prior art, and the reference is "Zonghua Zhang, Catherine E. Towers, and David P. Towers. Time efficient color fringe projection system for simultaneous 3D shape and color using optimum 3 - frequency selection. Optics Express, 2006, 14(14): 6444 - 6455".

[0045] Step 3: Collect the vertical deformed fringe pattern and the horizontal deformed fringe pattern;

[0046] Preferably, in step 3, the vertical sinusoidal straight fringe and the horizontal sinusoidal straight fringe obtained in step 2 are respectively modulated on the display screen 1, and the camera 3 collects the images transmitted through the transparent glass plate 2 to be measured, obtains the vertical deformed fringe pattern and the horizontal deformed fringe pattern, and stores them in the computer 4 for subsequent processing.

[0047] Step 4: Demodulate the vertical deformed fringe pattern and the horizontal deformed fringe pattern to obtain the vertical unfolded phase and the horizontal unfolded phase of the transparent glass plate 2 to be measured respectively;

[0048] Preferably, in step 4, the four - step phase - shifting method is used to demodulate the vertical deformed fringe pattern collected in step 3. One vertical folded phase pattern can be obtained from every four vertical deformed fringe patterns, and a total of three vertical folded phase patterns are obtained; then the optimal three - fringe selection method is used to process the three vertical folded phase patterns to demodulate the vertical unfolded phase of the transparent glass plate 2 to be measured;

[0049] The same processing is performed on the horizontal deformed fringe pattern to obtain the horizontal unfolded phase of the transparent glass plate 2 to be measured. Specifically: the four - step phase - shifting method is used to demodulate the horizontal deformed fringe pattern collected in step 3. One horizontal folded phase pattern can be obtained from every four horizontal deformed fringe patterns, and a total of three horizontal folded phase patterns are obtained; then the optimal three - fringe selection method is used to process the three horizontal folded phase patterns to demodulate the horizontal unfolded phase of the transparent glass plate 2 to be measured.

[0050] Preferably, in step 4, the best three-stripe selection method is a prior art, and the reference is "Zhang Z. Time-efficient color fringe projection system for simultaneous 3D shape and color using optimum 3-frequency selection. Optics Express, 2006, 14(14): 6444-6455".

[0051] Step 5: Obtain the surface scratch area and flatness error of the transparent glass plate 2 to be measured according to the vertical expanded phase and horizontal expanded phase obtained in step 4;

[0052] Preferably, in step 5, the specific method for obtaining the surface scratch area of the transparent glass plate 2 to be measured is as follows:

[0053] A5.1: Calculate the first-order derivative of the vertical expanded phase and horizontal expanded phase obtained in step 4 to obtain the phase gradient in the x direction and the phase gradient in the y direction, and then combine the gradient information in these two directions to obtain the phase gradient amplitude;

[0054] A5.2: Process the phase gradient amplitude using the Canny operator, and then perform binary processing on the processed phase gradient amplitude according to the accuracy requirement to set a threshold, and extract the scratch area;

[0055] A5.3: According to the scratch area, use an edge detection algorithm to further extract the contour of the scratch, and then perform morphological operations on the extracted scratch contour to optimize the scratch edge to obtain an optimized scratch edge;

[0056] Preferably, in step A5.3, the morphological operation consists of a dilation operation and an erosion operation performed in sequence.

[0057] A5.4: According to the optimized scratch edge, obtain the connected domain; then analyze the connected domain to determine the position of the scratch, and use the pixel counting method to calculate the number of lost pixel points in the scratch area;

[0058] A5.5: According to the actual physical size of the lost pixel points in each scratch area, convert the number of lost pixel points in the scratch area obtained in step A5.4 into the surface scratch area.

[0059] Preferably, in step 5, the specific method for obtaining the flatness error of the transparent glass plate 2 to be measured is as follows:

[0060] B5.1. Collect the deformation fringe pattern of a standard transparent glass plate that is the same size and shape as the transparent glass plate 2 to be measured and has high flatness, and then perform phase resolution to obtain the phase value Φ i , y j ) at the position (x ref (x i , y j );

[0061] B5.2. Calculate the root mean square error (RMSE) RMSE x in the x-direction and the root mean square error RMSE y in the y-direction of the transparent glass plate 2 to be measured;

[0062] Preferably, in step B5.2, the specific calculation of the root mean square error in the x-direction is: fix the position in the y-direction, i.e., y = y j , and calculate the root mean square error of the phase errors at all positions in the x-direction of the transparent glass plate 2 to be measured one by one as:

[0063]

[0064] In formula (1), Φ test (x i , y j ) represents the phase value of the transparent glass plate 2 to be measured at the position (x i , y j ); N x represents the total number of sampling points in the x-direction under the fixed y j ; x i is the position point in the x-direction, and y j is the fixed position in the y-direction;

[0065] The root mean square error RMSE x in the x-direction obtained from formula (1) is:

[0066]

[0067] In formula (2), N y represents the total number of sampling points in the y-direction under the fixed x i ;

[0068] Preferably, in step B5.2, the specific calculation of the root mean square error in the y-direction is: fix the position in the x-direction, i.e., x = x i , and calculate the root mean square error of the phase errors at all positions in the y-direction of the transparent glass plate 2 to be measured one by one as:

[0069]

[0070] In formula (3), Φtest (x i , y j ) represents the phase value of the transparent glass plate 2 to be measured at the position (x i , y j ); N y represents the total number of sampling points in the y direction with x fixed i ; y j is the position point in the y direction, and x i is the fixed position in the x direction.

[0071] The root mean square error RMSE in the y direction is obtained from Equation (3) y as follows:

[0072]

[0073] In Equation (4), N x represents the total number of sampling points in the x direction with y fixed j .

[0074] B5.3. Combine the root mean square error RMSE in the x direction x and the root mean square error RMSE in the y direction y to calculate the overall root mean square error RMSE of the transparent glass plate 2 to be measured total :

[0075]

[0076] B5.4. The overall root mean square error RMSE total is positively correlated with the flatness error. Use the overall root mean square error RMSE total to judge the flatness error.

[0077] Preferably, in step B5.4, the smaller the overall root mean square error RMSE total , the smaller the flatness error and the better the flatness.

[0078] Step 6. Combine the two indicators of the surface scratch area and the flatness error obtained in step 5 to judge whether the transparent glass plate 2 to be measured meets the requirements (i.e., is qualified): When both indicators meet their respective requirements, it is judged that the transparent glass plate 2 to be measured is qualified and meets the requirements; otherwise, it means that the transparent glass plate 2 to be measured is unqualified and belongs to defective products.

[0079] Preferably, in step 6, for the surface scratch area, compare the surface scratch area of the transparent glass plate 2 to be measured obtained in step 5 with the defect area threshold set according to the accuracy requirements. If the surface scratch area ≤ the defect area threshold, it is considered that the surface scratch area of the transparent glass plate 2 to be measured meets the requirements;

[0080] For the flatness error, compare the flatness error of the transparent glass plate 2 to be measured obtained in step 5 with the flatness error threshold set according to the precision requirements. If the flatness error ≤ the flatness error threshold, it is determined that the flatness error of the transparent glass plate 2 to be measured meets the requirements.

[0081] Example 1:

[0082] In step 1, the display screen 1 selected is a high-definition display screen produced by Weichensi Co., Ltd., with the model of Weichensi BM2, a resolution of 1920*1080 pixel, a size of 13.3 inches, and a video interface of mini-HDMI. The pixel distribution of this display screen is bar-shaped, with the characteristics of high flatness and high brightness, ensuring that the display screen shows high-quality stripe images.

[0083] The camera 3 uses an SVS-eco655CVGE color camera to collect images. This camera selects an ICX655AQA area array CCD type chip produced by Sony Corporation, which has the advantages of small distortion, high sensitivity, no afterimage, etc., and can collect complete image information within a large range at one time.

[0084] In step 2, the camera 3 is more sensitive to green light. Therefore, as Figure 3 shown, in this embodiment, four green vertical sine straight stripes with 64 stripes and a 90° phase shift from each other are displayed to avoid color difference to achieve better experimental results.

[0085] In step 4, from the three groups of horizontal deformed stripe images with different stripe frequencies collected in step 3, three horizontal folded phase diagrams are obtained, as Figure 4 shown.

[0086] In step 4, use the best three-strip selection method to process the three horizontal folded phase diagrams, and demodulate to obtain the horizontal unfolded phase diagram of the transparent glass plate 2 to be measured, as Figure 5 shown.

[0087] In step 5, A5.1. Calculate the first-order derivative of the vertical unfolded phase and the horizontal unfolded phase obtained in step 4 to obtain the phase gradient in the x direction and the phase gradient in the y direction. Combine the gradient information in these two directions to obtain a phase gradient amplitude diagram, as Figure 6 shown.

[0088] In step 5, A5.4. According to the optimized scratch edge, obtain the connected domain; then analyze the connected domain to determine the position of the scratch and obtain a scratch area diagram (as Figure 7 shown), and use the pixel counting method to calculate the number of lost pixel points in the scratch area. By Figure 7It can be seen that the red area is the connected region obtained according to the optimized scratch edge; the number of lost pixel points in the scratch area is 3422.

[0089] During the phase calculation process, the best effect is achieved when the pixel value occupied by a single fringe is 15-20 pixels. Therefore, the number of vertical fringes is selected as 64, 63, 56, and the number of horizontal fringes is selected as 49, 48, 42.

[0090] Matters not described in the present invention are applicable to the prior art.

Claims

1. A method for detecting surface defects of transparent objects based on transmissive structured light, characterized in that The method comprises the following steps: Step 1: Set up a measurement system; The measurement system includes a display screen (1), a transparent glass plate to be measured (2), a camera (3), and a computer (4); the computer (4) is communicatively connected to the display screen (1) and the camera (3) respectively; the display screen (1) is used to display the fringe image generated by the computer (4); the camera (3) is used to collect the deformed fringe pattern transmitted by the transparent glass plate to be measured (2); The optical centers of the display screen (1), the camera (3), and the transparent glass plate to be measured (2) are located on the same optical axis, and the transparent glass plate to be measured (2) is located between the display screen (1) and the camera (3); the deformed fringe pattern transmitted by the transparent glass plate to be measured (2) can be clearly imaged within the measurement field of the camera (3); Step 2: Generate a fringe image; Step 3: Collect a vertical deformed fringe pattern and a horizontal deformed fringe pattern; Step 4: Demodulate the vertical deformed fringe pattern and the horizontal deformed fringe pattern to obtain the vertical unwrapped phase and the horizontal unwrapped phase of the transparent glass plate to be measured (2) respectively; Step 5: Based on the vertical unwrapped phase and the horizontal unwrapped phase obtained in Step 4, obtain the surface scratch area and the flatness error of the transparent glass plate to be measured (2); Step 6: Comprehensively judge whether the transparent glass plate to be measured (2) meets the requirements based on the two indexes of the surface scratch area and the flatness error obtained in Step 5: when both indexes meet their respective requirements, it is determined that the transparent glass plate to be measured (2) is qualified and meets the requirements.

2. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 1, wherein In Step 1, the computer (4) is communicatively connected to the display screen (1) through a VGA interface and to the camera (3) through a gigabit network interface; In Step 1, the display screen (1) adopts an LCD display screen; the camera (3) adopts a CCD camera.

3. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 1, characterized in that, In Step 2, the computer (4) is used to generate three groups of vertical sinusoidal straight fringes and three groups of horizontal sinusoidal straight fringes that meet the requirements of four-step phase shifting and the selection of the best three fringes respectively. The number of fringes in the three groups of vertical sinusoidal straight fringes and the three groups of horizontal sinusoidal straight fringes both meet the selection of the best number of three fringes, and each group of vertical sinusoidal straight fringes and each group of horizontal sinusoidal straight fringes include four fringe patterns with a 90° phase shift between each other; In Step 3, the vertical sinusoidal straight fringes and the horizontal sinusoidal straight fringes obtained in Step 2 are respectively modulated on the display screen (1), and the camera (3) collects the image after transmission through the transparent glass plate to be measured (2) to obtain a vertical deformed fringe pattern and a horizontal deformed fringe pattern, which are stored in the computer (4) for subsequent processing.

4. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 1, wherein, In Step 4, the four-step phase shifting method is used to demodulate the vertical deformed fringe pattern collected in Step 3. One vertical folded phase map can be obtained from every four vertical deformed fringe patterns, and a total of three vertical folded phase maps are obtained; then the best three fringes selection method is used to process the three vertical folded phase maps to demodulate the vertical unwrapped phase of the transparent glass plate to be measured (2); The same processing is performed on the horizontal deformed fringe pattern to obtain the horizontal unwrapped phase of the transparent glass plate to be measured (2).

5. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 1, characterized in that, In Step 5, the specific method for obtaining the surface scratch area of the transparent glass plate to be measured (2) is: A5.

1. Calculate the first-order derivatives of the vertically and horizontally unwrapped phases obtained in step 4 to obtain the phase gradients in the x and y directions, and then combine the gradient information in these two directions to obtain the phase gradient magnitude. A5.

2. Process the phase gradient magnitude using the Canny operator, and then perform binarization on the processed phase gradient magnitude according to the accuracy requirements by setting a threshold to extract the scratch area. A5.

3. According to the scratch area, use an edge detection algorithm to further extract the contour of the scratch, and then perform morphological operations on the extracted scratch contour to optimize the scratch edge to obtain the optimized scratch edge. A5.

4. According to the optimized scratch edge, obtain the connected regions; then analyze the connected regions to determine the position of the scratch, and use the pixel counting method to calculate the number of lost pixel points in the scratch area. A5.

5. According to the actual physical size of the lost pixel points in each scratch area, convert the number of lost pixel points in the scratch area obtained in step A5.4 into the area of the surface scratch.

6. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 5, wherein In step A5.3, the morphological operation consists of a dilation operation and an erosion operation performed in sequence.

7. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 1, wherein In step 5, the specific method for obtaining the flatness error of the transparent glass plate (2) to be measured is as follows: B5.

1. Collect the deformation fringe pattern of a standard transparent glass plate that has the same size and shape as the transparent glass plate (2) to be measured and has high flatness, and then perform phase resolution to obtain the phase value Φ i , y j ) at the position (x ref (x i , y j ). B5.

2. Calculate the root mean square error RMSE of the transparent glass plate (2) to be measured in the x direction x and the root mean square error RMSE in the y direction y ; B5.

3. Combine the root mean square error RMSE in the x direction x and the root mean square error RMSE in the y direction y to calculate the overall root mean square error RMSE of the transparent glass plate (2) to be measured total : B5.

4. Overall Root Mean Square Error RMSE total There is a positive correlation with flatness error, and the overall root mean square error RMSE is used total to judge the flatness error.

8. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 7, characterized in that, Preferably, in step B5.2, the calculation of the root mean square error in the x direction is specifically as follows: fix the position in the y direction as y = y j , and calculate the root mean square error of the phase errors at all positions in the x direction of the transparent glass plate (2) to be measured one by one as: In formula (1), Φ test (x i , y j ) represents the phase value of the transparent glass plate (2) to be measured at the position (x i , y j ); N x represents the total number of sampling points in the x direction with y j fixed; x i is the position point in the x direction, and y j is the fixed position in the y direction. The root mean square error RMSE in the x direction is obtained from Equation (1). x It is as follows: In formula (2), N y represents the total number of sampling points in the y direction with x fixed i as described below.

9. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 7, characterized in that, In step B5.2, the specific calculation of the root mean square error in the y direction is as follows: Fix the position in the x direction at x = x i , and calculate the root mean square error of the phase errors at all positions in the y direction of the transparent glass plate (2) to be measured one by one as: In Equation (3), Φ test (x i , y j ) represents the phase value of the transparent glass plate (2) to be measured at the position (x i , y j ); N y represents the total number of sampling points in the y direction with x fixed i ; y j is the position point in the y direction, and x i is the fixed position in the x direction. The root mean square error RMSE in the y direction is obtained from Equation (3). y It is as follows: In formula (4), N x represents the total number of sampling points in the x direction at a fixed y j value.

10. The method for detecting surface defects of a transparent object based on transmissive structured light according to claim 1, wherein In step 6, for the surface scratch area, compare the surface scratch area of the transparent glass plate (2) to be measured obtained in step 5 with the defect area threshold set according to the accuracy requirements. If the surface scratch area ≤ the defect area threshold, it is determined that the surface scratch area of the transparent glass plate (2) to be measured meets the requirements. For the flatness error, compare the flatness error of the transparent glass plate (2) to be measured obtained in step 5 with the flatness error threshold set according to the accuracy requirements. If the flatness error ≤ the flatness error threshold, it is determined that the flatness error of the transparent glass plate (2) to be measured meets the requirements.

Images